1 /*============================================================================= 2 Copyright (c) 2001-2011 Joel de Guzman 3 4 Distributed under the Boost Software License, Version 1.0. (See accompanying 5 file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt) 6 ==============================================================================*/ 7 #if !defined(BOOST_SPIRIT_UC_TYPES_NOVEMBER_23_2008_0840PM) 8 #define BOOST_SPIRIT_UC_TYPES_NOVEMBER_23_2008_0840PM 9 10 #if defined(_MSC_VER) 11 #pragma once 12 #endif 13 14 #include <boost/cstdint.hpp> 15 #include <boost/foreach.hpp> 16 #include <boost/regex/pending/unicode_iterator.hpp> 17 #include <boost/type_traits/make_unsigned.hpp> 18 #include <string> 19 20 namespace boost { namespace spirit 21 { 22 typedef ::boost::uint32_t ucs4_char; 23 typedef char utf8_char; 24 typedef std::basic_string<ucs4_char> ucs4_string; 25 typedef std::basic_string<utf8_char> utf8_string; 26 27 template <typename Char> to_utf8(Char value)28 inline utf8_string to_utf8(Char value) 29 { 30 // always store as UTF8 31 utf8_string result; 32 typedef std::back_insert_iterator<utf8_string> insert_iter; 33 insert_iter out_iter(result); 34 utf8_output_iterator<insert_iter> utf8_iter(out_iter); 35 typedef typename make_unsigned<Char>::type UChar; 36 *utf8_iter = (UChar)value; 37 return result; 38 } 39 40 template <typename Char> to_utf8(Char const * str)41 inline utf8_string to_utf8(Char const* str) 42 { 43 // always store as UTF8 44 utf8_string result; 45 typedef std::back_insert_iterator<utf8_string> insert_iter; 46 insert_iter out_iter(result); 47 utf8_output_iterator<insert_iter> utf8_iter(out_iter); 48 typedef typename make_unsigned<Char>::type UChar; 49 while (*str) 50 *utf8_iter++ = (UChar)*str++; 51 return result; 52 } 53 54 template <typename Char, typename Traits, typename Allocator> 55 inline utf8_string to_utf8(std::basic_string<Char,Traits,Allocator> const & str)56 to_utf8(std::basic_string<Char, Traits, Allocator> const& str) 57 { 58 // always store as UTF8 59 utf8_string result; 60 typedef std::back_insert_iterator<utf8_string> insert_iter; 61 insert_iter out_iter(result); 62 utf8_output_iterator<insert_iter> utf8_iter(out_iter); 63 typedef typename make_unsigned<Char>::type UChar; 64 BOOST_FOREACH(Char ch, str) 65 { 66 *utf8_iter++ = (UChar)ch; 67 } 68 return result; 69 } 70 71 // Assume wchar_t content is UTF-16 on MSVC, or mingw/wineg++ with -fshort-wchar 72 #if defined(_MSC_VER) || defined(__SIZEOF_WCHAR_T__) && __SIZEOF_WCHAR_T__ == 2 to_utf8(wchar_t value)73 inline utf8_string to_utf8(wchar_t value) 74 { 75 utf8_string result; 76 typedef std::back_insert_iterator<utf8_string> insert_iter; 77 insert_iter out_iter(result); 78 utf8_output_iterator<insert_iter> utf8_iter(out_iter); 79 80 u16_to_u32_iterator<wchar_t const*, ucs4_char> ucs4_iter(&value); 81 *utf8_iter++ = *ucs4_iter; 82 83 return result; 84 } 85 to_utf8(wchar_t const * str)86 inline utf8_string to_utf8(wchar_t const* str) 87 { 88 utf8_string result; 89 typedef std::back_insert_iterator<utf8_string> insert_iter; 90 insert_iter out_iter(result); 91 utf8_output_iterator<insert_iter> utf8_iter(out_iter); 92 93 u16_to_u32_iterator<wchar_t const*, ucs4_char> ucs4_iter(str); 94 for (ucs4_char c; (c = *ucs4_iter) != ucs4_char(); ++ucs4_iter) { 95 *utf8_iter++ = c; 96 } 97 98 return result; 99 } 100 101 template <typename Traits, typename Allocator> 102 inline utf8_string to_utf8(std::basic_string<wchar_t,Traits,Allocator> const & str)103 to_utf8(std::basic_string<wchar_t, Traits, Allocator> const& str) 104 { 105 return to_utf8(str.c_str()); 106 } 107 #endif 108 }} 109 110 #endif 111